Thermal sensor issue i7-7700k?

[CK7]

I have a brand new build; EVERYTHING NEW. i7-7700k is running at stock speeds. I have the RAM set to XMP for DDR4-2666. Motherboard is Asus Maximus IX Hero Z270.

I have found that the i7-7700k reports a momentary (a second or less) temperature spike +25 > 35 degrees Celsius anytime a program is opened, a webpage is opened, a background app runs etc. The temperature blip cascades through the cores in random order; not the same every time. This causes my heatsink fan to constantly cycle up and down. Temperatures otherwise report as steady, normal increases. Peak temperature under Prime95 blend test is 71 degrees Celsius.

Attempted solutions:

I have re-installed my heatsink and thermal paste with no change.

I have tried to manually set my fan speed in the bios. The only setting that avoids this issue is setting the temperature / fan at a constant (and loud) 80-100%. I've tried PWM and DC mode.

I have found a few user reports elsewhere on the web, all reasoning that it's just the way it is. I don't accept that. Opening a folder or browser should not spike temps +30 degrees. Not only is the fan cycling annoying, it puts undue stress on my fan; possibly shortening its lifespan.

What's the answer, if any? RMA?

[TGrab1]

Well your is much more smooth than mine. I however use offset mode for my stock config. I only run my overclocked profiles if I need to squeeze out a few more fps or for bragging rights :-D.

Anyways you will see mine produces a much more ringing bell effect. Also note the highest my Vcore hits is 1.18 at 4525mhz.

So the dips for me push me well below what most cpus would probably be stable. Maybe I just have a great sample as I am 100% p95 stable a long with IBT, occt, aida64, and realbench.

here are my temps for the test.

I saw a spike to about 54C, but the majority of the time I was in the mid 40's. temps didnt even get high enough for my pump/fans to even increase their speeds. Which make me curious what my temps would have been if I had my system running at 100% fans/pump.

Anyways if this isn't the test you wanted let me know and I can run another. As you can see though I get a pretty good amount of fluctuation compared to your manual settings. I was not running LLC though.

[idata]

Thanks T. What was the CPU offset voltage?

What I'm interested in are the dips, like the one you got on OCCT start (about 60mv from the top value).

10 mins is really too short. You would need to run the test for at least 30mins, but preferably an hour ... you might then see dips during the test (like the start dip). What I found is that these dips sometimes (but not always) cause the test to fail (as you would expect) if your offset voltage is too low.

Ideally it would be better to set the voltage at manual, with LLC mid-way. The frequency doesn't matter - I see these blips at 4.2GHz as well as at 4.8GHz.

The thing is that the dips don't really matter at low frequencies because I can just up the voltage a bit so that the dips never dip too low. However at 4.8 or above they really do matter because every mv counts in terms of heat (and getting too close to the killing zone, so to speak).

So if you do have the time, if you could run the test again for 30mins or an hour with the system set to, say:

4.2GHz

Manual 1.2v (say)

LLC to mid-way (but at any rate not Auto or off)

C states, speedstep disabled

... it will be very interesting to see what your Vcore looks like.

Cheers

Robert

[TGrab1]

So this test. 30 minutes.

4.8ghz at manual 1.285

LLC at 4

Speed shift - off

Speed step - off

C-state off

Max temp was 66C at one point, but the temp was in the mid to low 50's 95% of the time.

The one thing I am noticing.. is even with LLC on it is making sure to stay below my manual setting.. some of this is due to vdroop, but I would expect to see some spikes over.. but that didn't happen.

I do have a few spikes (4) that are at a pretty good drop, which I do not have an explanation for off hand but they all dropped to about 1.235 despite a setting of 1.285 manual vcore.

That being said a higher LLC might solve that issue...

The thing is I do not have stability issues over these blips. I ran this OC under p95 for 20-30 minutes prior to this without issue.. so yes that isn't a long p95 test, but a unstable OC will normally throw an error in p95 in that period.

I guess one different between me and you though is in thermal headroom. I can afford to go a little higher in Vcore to make sure the blips aren't an issue due to being delidded and on custom water. So even at the settings being use I still have 20C or so of head room before I break 80c going off my highest spikes. 30C if you go off of my average 50-55C temps.

[idata]

Many thanks T ... that's exactly the test I was looking for.

You're able to get a stable OC at 1.285 whereas I'm having to go to 1.32, which just shows the effect of your delid and custom cooling ... AND you have plenty of thermal headroom whereas I do not. So without at a delid, I really can't go much above 4.8, or at least not for any applications that are power-hungry.

But what is very interesting is that you are also getting the blips.

If you look at the end of the test, you can see that the peak voltage at idle is about 1.280, which absolutely correct with your LLC setting: a value of 4 is quite low (for ASUS) and you shouldn't be going over the offset voltage of 1.285. So the voltage is varying between 1.265 and 1.280 which looks good.

Under load the voltage drops to between 1.250 and 1.265, which is what you would expect with a low LLC value. Again this looks absolutely fine. What is not fine are the blips down to 1.232: exactly the same thing I'm getting on my current board and also on the one I've replaced. The test passed because 1.232 may be still OK, just, or you were just lucky. You may find that if you run the test overnight that one of these blips will fail the test.

So either these blips are a 'feature' of these ASUS motherboards, or there's an explanation for them which I'm missing. They do look like heavy ringing, but then you would expect to also see the opposite, high voltage blips, and these are not there. We would really need to use an oscilloscope to really see what's going on. At any rate, I do think they are limiting our stable overclock.

P95 doesn't show these blips, I don't think, probably because P95 is unrelenting and doesn't vary the load to the same extent (possibly the mixed test does). So for this issue at least, OCCT is a more demanding test than P95 small FFT.

What would really be MOST interesting would be for someone else to try this test on a non-ASUS board. Any volunteers?

Cheers,

Robert

[EWerd]

HI Intel! Hans Jürgen Werner Publicly mislead us?

https://youtu.be/80xb9fHio4g?t=5m48s Что Intel думает о RyZEN? Эксклюзивное интервью с директором Intel - YouTube

[jstan141]

Hi,

Can I buy i7 7700K without fear? Its temperature can be dangerous?

[TGrab1]

Robert,

It could also just be this board. My board and your board are the exact same board with different video connections built in.

I didn't go with a high-end board as the mid range boards have always overclocked plenty well for me, but that could be why we are seeing those blips. Could just be the quality of the VRM used in this model. I would also be interested in seeing what it looks like on a different board.

If I have time tonight I can run a very similar test on a 2600K machine with a decent overclock... Different cpu, but maybe the vcore portion will prove useful.

I am also pretty confident that the overclock I had would not fail an overnight test... If I can get a OC to pass a 30 min P95 then 99% of the time it will be stable for anything else I can throw at it. I have been able to hit 5.2ghz so far at about 1.365 (manual) which was just to see if I could do it.. and it did survive a overnight p95 test. I probably could get the voltage down more if I used adaptive and LLC, but with Vdroop is was only pulling about 1.34x under load anyways.

[idata]

Could you report this to ASUS please? The support guy I'm in touch with (his name is Paul) said that the issue is not normal and appears to be a fault; he also said that it has not been reported before. He recommended a replacement of the board. He said that the Prime motherboards are entry-level boards and not intended for extreme overclocking (contrary to their literature; and I also showed him the issue with the board at 4.2GHz).

Now we're seen this on 3 Prime-Z270 boards, so it seems to be somewhat of a feature of these boards. I've asked ASUS "In the meantime I would appreciate an explanation for what these blips are from ASUS. Either they are caused by a fault on these motherboards, or there's a simple explanation for them and your R&D department should be able to provide it without breaking any sweat." If you also report the issue then it will reinforce the support request.

It would be interested to see what your 2600K looks like.

With any luck someone on this topic with a ROG or TUF board or with an MSI or other board will give this test a try. Please guys!

Here are the test requirements:

Set Vin to Manual 1.25v at 4.5GHz, say

Set LLC to mid-way setting

Turn off C-States, Speedstep and Speedshift

Run OCCT with 5 mins pause at start and end, for at least 30mins

Post Vcore graph on this thread

Cheers

Robert

[TGrab1]

Well if they are claiming this is a low end board the ROG will be no different as they were literally the same price. I went with the Prime due to better onboard audio and usb C and 3.1 support.

[AElib]

I Have Asus ROG Maximus VIII Extreme would that help, or are you talking about Z270 Board?

[idata]

Hi ZeoxZariX ... yes, if you could run this test it would be very helpful! ASUS can hardly claim that your board is not aimed at overclockers, and if you get the same blips then it shows that the problem is not restricted to the Prime series of boards ... on the other hand, if you do not get the blips then it shows that there is in fact an issue with the Prime motherboards.

Looking forward to seeing your results!

Cheers

Robert

[AElib]

I will be home in one hour, i will post as soon as possible.

The ROG Maximus VIII Extreme it has been built for Extreme OC, let's see what they would say.

[AElib]

Set Vin to Manual

What's Vin? :-(

[AElib]

At the moment i don't have Internet at home because i moved, i took a pic with my phone.

I hope this is what you wanted.

[idata]

Hi ... exactly what I want

But I'm puzzled by your VCore values. Do you have the CPU Core/Cache Voltage set to Manual? I don't understand how your Vcore is so low. Also, did you set the Load Line Calibration to 4 or 5?

At any rate, what is interesting is that you are getting no negative spikes during the test.

If you could redo the test for 30mins minimum, with the CPU Core/Cache Voltage set to Manual Mode, Additional Turbo Mode CPU Core Voltage (I think that's what it's called in your BIOS) to 1.25, say, Offset Voltage to 0, Load Line Calibration to 5, that would be really great!

Robert

[TGrab1]

Ya those Vcore values are super low. Either that is the best golden chip I have ever seen or something else is going on here.

[AElib]

Yes the CPU Core/Cache Voltage were set to Manual and Load Line Calibration was set at 5, i dont understand why my Vcore was so low.

So, here is the second test you ask for.

CPU Core/Cache Voltage set to Manual Mode, Additional Turbo Mode CPU Core Voltage to 1.25, Offset Voltage to 0, Load Line Calibration to 5,

s

[AElib]

The same low voltages in the second test, i dont know what is happening.

[TGrab1]

The only difference I can see is the version of OCCT you are running. Maybe try updating to 4.5.0. Might be an issue with the chipset being newer than that version.

What will be interesting is if you update to the new version and start seeing those blips. Then we might be able to narrow this down to the software.

[AElib]

I was thinking the same thing, i wont have internet untill tomorrow but i will go to a internet point and download OCCT 4.5.0

[idata]

The voltages seem to be scaled to 50%, probably the OCCT version as TGrable suggested. You could run HWInfo64 and graph VCore to get a second opinion, so to speak.

But apart from the scaling, it looks very good. If you can it would be great to run the test for an hour, just to be sure these spikes are definitely not there on your board.

ASUS support are being singularly unhelpful ... they won't run any tests or answer any questions: the support guy just tells me to get the board replaced (easy for him to say that, he doesn't have to do the work!). I'm going to escalate this issue as best I can.

Cheers

Robert